Background
Hazardous alcohol use is an important contributing factor to morbidity and mortality worldwide, with social- and economic costs [
1‐
3]. According to studies using screening questionnaires, 21–34% of patients treated in intensive care units (ICUs) reported a history of hazardous alcohol use [
4‐
8]. A history of hazardous alcohol use is thus common among patients admitted to the ICU [
4‐
9].
Screening for alcohol use and performing brief interventions (BI) to motivate patients to reduce hazardous alcohol consumption have been reported to be effective strategies in primary health care and hospital settings [
10‐
15] Patients with a history of hazardous alcohol use may be more receptive to learning about the risks of alcohol use during hospitalization [
16,
17]. The first step in the treatment of patients suspected of hazardous alcohol consumption is screening, typically using validated questionnaires. A positive screening result leads to a BI which usually consists of six key elements: feedback, responsibility, advice, menu for change, empathy and enhancing self-efficacy (summarized by the acronym FRAMES) [
10,
18]. In addition, the BI may include provision of self-help materials and contact information on additional support available [
19].
The period after a severe illness may be a teachable moment, suitable for providing information and support and for motivating patients to reduce their alcohol use. In this study, we aimed to examine the effectiveness of a BI compared to treatment as usual (TAU) in reducing hazardous alcohol use in ICU survivors. Our objective was to compare the alcohol intake between patients receiving BI with those with TAU, at 6 and 12 months after randomization. We hypothesized that BI would lead to decreased alcohol consumption compared to TAU, 6 and 12 months after randomization.
We also compared the change in AUDIT-C scores from baseline, health-related quality of life (HRQoL), and survival at 12 months.
Discussion
We conducted a randomized, controlled multi-center study to assess whether a BI at the time of ICU discharge affected alcohol consumption 6 and 12 months later in patients with a history of hazardous alcohol use. To the best of our knowledge, this is the first randomized controlled trial to examine the effect of a BI on alcohol consumption after critical illness, and the first study to attempt to study the effectiveness of a BI in general ICU patients with a history of hazardous alcohol consumption. Because the trial was terminated early it lacks statistical power to reject or confirm the hypothesis of a 10% reduction in alcohol consumption with a BI delivered early after ICU discharge, compared to treatment as usual. On the other hand, no trend was observed that the BI would have decreased the alcohol consumption more than TAU. An important finding of our study was that regardless of the group allocation, considerable number of participants in each group reported no use of alcohol during the week preceding the 6- and 12-month follow-up assessments. Six months after randomization, 4% of individuals in the BI group and 11% of those in the TAU group reported total abstinence. At the 12-month follow-up, the corresponding proportions were 10% and 15% in the BI and TAU groups, respectively. Surprisingly, the prevalence of reported abstinence had increased in both groups at the 12-month follow up despite two-thirds of the participants being alcohol dependent at baseline. Screening for hazardous alcohol consumption and BIs are recommended in all healthcare encounters by WHO and Finnish Current Practice Guidelines [
19,
32] and ICU survivors should receive support in reducing their alcohol consumption to avoid further negative health consequences. Our results suggest that critical illness is a strong motivational factor for reducing alcohol consumption.
The effectiveness of BIs has been demonstrated in primary care settings [
11,
13,
33]. There is conflicting evidence on the effectiveness of BIs in emergency settings [
12,
34], with some studies finding that BIs were not effective in emergency settings [
35,
36], but another study finding that alcohol-related injury may increase the effectiveness of BI [
37]. The usefulness of BIs may depend on the target population. For example, BIs in emergency settings were found not to be useful for patients with substance abuse, admitted because of violence-related events [
36]. However, alcohol interventions to trauma patients were associated with reduced alcohol-related injuries [
38,
39]. The results of studies on the effectiveness of BIs for hazardous alcohol use in hospitalized patients are conflicting [
10,
12,
39‐
42]. We observed no trend to more decreased alcohol consumption after BI than TAU, but AUDIT-C scores decreased significantly in both the BI and TAU groups. The BI was delivered in the ICU or shortly after ICU discharge in the hospital ward. The timing was soon after critical illness and perhaps not optimal for a BI. Possibly, the experience of critical illness, participating in the study, and answering AUDIT questionnaire rather than the BI may have motivated the patients (in both groups) to reduce hazardous alcohol use. According to the World Health Organization (WHO), hazardous alcohol use is defined as over 20 g per day for women, and over 40 g per day for men [
43]. In our study, only a few patients reported consuming more alcohol than the WHO recommendation at the 6- and 12-months follow-ups, but according to AUDIT-C scores from the same time point hazardous use was still prevalent. This discrepancy is likely explained by irregular drinking, a common drinking pattern in Finland [
44].
A desire to change drinking habits is thought to be the main component of a successful BI [
45]. Willingness to change is an important part of motivation and readiness to change [
45]. We found no difference between the groups in their willingness to change alcohol use habits or the participants´ confidence in their ability to change, at the follow-up assessments. A previous study reported that willingness and motivation to change alcohol use habits were associated with successful treatment outcomes for hazardous alcohol use [
46‐
48]. In addition, the confidence in one’s ability to change is important in reducing harmful alcohol consumption [
45]. In our study, the participants’ confidence in their ability to change was high in both groups at both follow-ups. In both groups, we found a moderate correlation between both willingness to change and confidence in the ability to change at the 6-month follow-up and the change in AUDIT-C scores from 6 to 12 months.
Although 12-month mortality was low in our population, the baseline AUDIT score was higher in non-survivors than survivors. We found no difference in crude or adjusted mortality between the intervention and control groups. This is contrary to the results of a previous meta-analysis, where a BI associated with decreased mortality [
10,
15]. In our study, there were no major between-group differences in the quality of life at the 12-month follow-up although the patients in the BI group reported coping self-care better than those in the control group.
Our study has several strengths. It was a multicenter study and included both emergency medical and surgical patients’ admissions, which increases the generalizability of our results. We used central randomization, with stratification for center, sex, and dependency, and delivered the intervention in a structured manner using trained investigators and recommended interview techniques. In addition, the outcome assessors were blinded to the group allocation.
Our study has limitations. Patient recruitment was terminated early due to the slow recruitment during the pandemic. As a result, we recruited only 39% of the calculated sample size. The loss to follow-up rate was higher than the estimated 25%, which is a recognized source of bias in studies in similar populations [
10‐
12]. We did not analyze the results separately in women and men. Due to the nature of the intervention, blinding was not feasible in delivering the intervention. Although not routinely available, patients in both groups may have received support for reducing alcohol consumption, both during and after hospitalization. AUDIT interview at baseline with both groups (for stratification purposes) may be perceived as a type of support. The primary endpoint was based on self-report, involving the inherent weaknesses of self-report, and risk of response and recall bias. Our primary endpoint probably reflects short-term alcohol consumption, especially in subjects with alcohol addiction, supported by our finding of different prevalence of hazardous use when using the AUDIT-C and recent intake as methods of assessment. Both belong, however, to the recently published core outcome set for BI trials [
49]. The collection of data for the primary endpoint using two alternative methods may have caused bias, and despite efforts there were missing data. Finally, we cannot exclude selection bias regarding the screening process not covering consecutive patients, patients´ willingness to participate, and the response rate (e.g., favoring participation of motivated patients and responses from patients with successful reduction of alcohol use).
Despite these limitations, our results are important. Reducing hazardous alcohol consumption is crucial to decrease the negative health consequences associated with hazardous alcohol use. Studies focusing on optimal timing, number and type of intervention, and objective method for screening and assessment of outcome in ICU survivors with history of hazardous alcohol use are called for. A follow-up clinic two to three months after hospital discharge might be a suitable setting for a BI intervention trial, as our results point out that the ICU survivors are motivated to change their alcohol use habits. In future studies, the sample size calculation should consider a higher attrition rate and make all efforts to reduce missing data. Outcomes collected should align with the published core outcome set [
45].
Acknowledgements
We thank Tarja Suhonen, Minttu Saario, Elina Lappi, Leena Pettilä, Jonna Heinonen, Sari Sutinen, Vuokko Korpilahti, Eija Koponen, Eeva-Liisa Piiparinen, Kirsi Pulkkinen, Sanna Räsänen, Päivi Kainulainen, Marjo Tauren, Anna-Maria Kuivalainen, Päivi Haltia, Jussi Kauppinen, Satu Kentala, Eveliina Loikas, Marita Suna, Anna Hellström, Eljas Honka-Hallila, Katri Holmström, Tiina Koivisto, Anne Mäki, Anne Mäkimaa, Anna Packalen, Arja Roms, Tuula Sorvari, Tarja Viitala, Helka Tervas, Auli Palmu, Sanna Ristimäki, Sanna Mäkinen, Kaija Veijalainen, Iiris Nieminen, Anna Wootten, Satu Helminen and Merja Hietanen for delivering BI and data acquisition. We thank Olavi Koivisto for his invaluable help with statistical analysis.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.